So Wikipedia has a handy chart for handicapping chess games.
However I cannot seem to find one relating to both players Elo rating. For instance one handicap is a knight removed, what should the Elo rating difference be for this to be fair?
I’d venture to guess between higher level players the range would be much smaller. For my friends and I we’re looking at 1100 to 1600.
[Moderating]
You accidentally posted this in GD. I moved it for you.
Your link does mention Elo-rating equivalency with a cite from a Grandmaster posting on an online forum:
[T]he Elo equivalent of a given handicap degrades as you go down the scale. A knight seems to be worth around a thousand points when the “weak” player is around IM level, but it drops as you go down. For example, I’m about 2400 and I’ve played tons of knight odds games with students, and I would put the break-even point (for untimed but reasonably quick games) with me at around 1800, so maybe a 600 value at this level. An 1800 can probably give knight odds to a 1400, a 1400 to an 1100, an 1100 to a 900, etc. This is pretty obviously the way it must work, because the weaker the players are, the more likely the weaker one is to blunder a piece or more. When you get down to the level of the average 8 year old player, knight odds is just a slight edge, maybe 50 points or so.
Handicap (chess) - Wikipedia
This indicates to me that since handicapping really isn’t a thing anymore (according to that same article) and was never really formal, there is no such conversion table available outside of some people’s personal conjecture.
It is actually the opposite of this. The higher the Elo rating of the non-handicapped player, the more devastating the effect of giving them a knight/rook/pawn/whatever advantage, because they are less likely to make blunders which erase the difference in material, and they are better at converting small advantages into a won endgame.
Almost six years ago, a series of 6-game matches were played between human grandmasters and one of the leading computer chess programs at that time, with the humans receiving different types of handicaps in each game. From their (albeit very limited) dataset, they estimated the following Elo-equivalents for these handicaps, assuming the non-handicapped player is at a grandmaster level:
| Odds | Description | Approximate Elo value |
|---|---|---|
| Queen odds | Remove Qd1 | 2000 |
| Rook odds | Remove Ra1 | 1400 |
| Knight odds | Remove Nb1 | 1100 |
| Two-pawn odds | Remove two pawns | 400-700 |
| Exchange odds | Remove Ra1 and Nb8 | 500 |
| Pawn and two moves | Remove f7, White plays two moves to start | 600 |
| Pawn and move | Remove f7 | 450 |
Yeah, I would definitely intuit that the higher the level, the MORE advantage a little bit of material is.
I wonder if anyone has done a trial simply involving a computer playing against itself against various levels to get some estimates? I’d guess you’d get a pretty good ballpark figure and some interesting data visualizations out of that.
My friends know how to play, but haven’t studied. I am around 1500 on Chess.com and SocialChess so loss of black queen side knight seems fair. We’ll see how it goes.
I realize my wiki cite had some of the information I asked for, but it doesn’t seem complete, so I was a bit confused.
A curious thing with piece odds is that the meta strategy of the game is changed from move 1. At the levels you mention (1100 vs 1600 or so), the weaker player is unlikely to mount a successful middle game attack against the stronger player even down a piece, so they are incentivized just to develop quickly (but safely) and to trade things early and often to simplify to a piece-up endgame. In turn, the stronger player who is down a piece should play into lines that make trades difficult while engineering complications where tactical superiority may shine through.
If both players are aware of this meta, it makes for a game that feels very different from regular chess, at least for the first 15 moves or so. Still lots of fun, of course! But very different. Playing with time odds feels more like normal, and is also more readily “tunable”.
There’s also the issue that sometimes a piece is a liability as well as an asset. With a pawn handicap, for instance, the player who’s down a pawn will then have an easier time developing whatever was behind that pawn.
Well, not necessarily whatever was behind (at least directly behind) that pawn. It depends on what piece is there. Take away an a or h pawn then, yeah, those rooks can come flying out right away. Take away the f pawn and – well – I guess you’ve freed up a king a file over that you probably don’t want to move. And that’s exactly why with a pawn handicap, that’s the pawn that gets taken off the board.
I do wonder if anyone has done any analysis of whether taking away certain pawns is actually an advantage and how much of one it is to the player giving up the pawn. I suspect in multiple cases it may confer an advantage.
My favorite story about odds play was the one related about Alekhine:
Alekhine was on a long train journey and a fellow traveller asked him if he would like to pass the time with a game of chess. So Alekhine set them up and removed one of his Rooks to make it more of a challenge. The fellow traveller was indignant: “What do you mean by giving me rook odds? You don’t even know who I am?”. Alekhine replied “If I couldn’t give you rook odds, I would know who you were.”
Ha! That gave me a chuckle.
It would never be an advantage. The only plus is a tiny bit more freedom of movement, but you get plenty of freedom of movement just advancing a pawn rather than removing it entirely, and that pawn controls squares (limits opponent’s movement), gives support for your developing pieces, and is a full extra pawn for an endgame. You do have to spend a tempo moving said pawn, but that’s way worth it.
From what very basic checking I did with Lichess.org’s analysis (Stockfish 13+ engine), with white going first and white being the one to give up the pawn, I get these numbers: (A positive number indicates White advantage; a negative number black advantage)
Base position (no pawns removed): +0.2
a pawn removed: -0.5
b pawn removed: -1.3
c pawn removed: -0.9
d pawn removed: -1.2
e pawn removed: -1.1
f pawn removed: -1.5
g pawn removed: -1.6
h pawn removed: -0.5
So removing the rook pawns only confers about a half pawn advantage to black using this rudimentary analysis. When I look at its analysis game, though, I see typical opening book stuff. I’m wondering if there is something out there that is outside the standard opening books that freeing up an early rook might take advantage of. I’d need to somehow turn the opening book off (which I don’t see an option to do) or use something like a self-learning algorithm like Leela to be completely convinced there isn’t the possibility of an immediate threat conferring an advantage. I’m leaning towards no now, but I’d like to see more.
Still it is interesting how much the choice of pawn varies the disadvantage.
ETA: Other interesting things I see from various analyses. If I get rid of both rook pawns, it’s still only a -0.9 disadvantage to white, meaning getting rid of the two rook pawns is only as much a disadvantage as getting rid of any single other pawn (and, in most cases, even less a disadvantage.)
ETA2: Actually, as I let it play even farther down the depth, I’m now getting a -1.6 disadvantage at depth 34 vs the -0.9 at depth 24 that was the default. Still, less than two full pawns and about as much a handicap as just removing the g pawn.
A rook pawn is in some sense only “half a pawn” anyway in the early game because it only controls one file instead of two. I suspect that’s the main reason the low- to mid-depth evaluation is so much different for rook pawns vs. non-rook pawns, more so than for the possibility of a rook lift.
As for rook mobility, one can always just play 1. a4 xxx 2. Ra3 and still get a rook out early without losing a pawn. A rook developed via a deleted pawn only gets to two additional ranks that are very unstable places for a rook to be since it can get attacked with tempo every time black plays out a pawn of his own and improves his own development. In other words, playing 1. Ra4 or 1. Ra5 costs white at least as much development time as it gains. (E.g., with no pawn on a2 to start, where would you put the rook after 1. Ra4 b5 such that it is doing something useful and not getting immediately attacked again on black’s next move?)
Aside: Stockfish’s evaluation is not dependent on the opening book, so there’s nothing to turn off. If you play against an engine, there could be a toggle to allow it to use an opening book or not, but the evaluation function is just the evaluation function.
To add: The reason you see normal opening looking stuff when a rook pawn is deleted is because 1. e4 and 1. d4 (etc.) are still good for all the same reasons as usual whether there’s a pawn on a2 or not. Moving the rook out is just a bad move in that spot.
Yeah, I mean, you don’t usually want to get those types of pieces out there so early as they can get chased around, but I was wondering if there may be some sort of early combo that is possible with one of the pieces unencumbered to start with. It seems not.
Sure, you can move a rook out early by moving its pawn to 4, but then you still have to zigzag around it. To reach, say, one of the four center squares that way would take at least four moves (one pawn move and three rook moves, or two of each), as opposed to two moves if the pawn is completely absent.
Though it’s also true that a rook that far forward in the early game could be chased around a lot by enemy pawns. And using a valuable piece like a rook in the opening maneuvers to control the center is tricky, because you have to make absolutely sure that the rook is the last piece to move in any trades (i.e., the one making the capture that won’t be counter-captured).
That’s interesting, because games played at pawn odds (whether White or Black surrenders the pawn), traditionally offered the f pawn, because it was thought that surrendering that pawn yielded the least compensating advantage. Looks like Philidor, Morphy, et al., may have been finagling just a tiny bit.
The argument over what is White’s worst opening move (in a normal, no-odds game) generally came down to 1. f3 vs. 1. g4 (note that the “fool’s mate” involves both these moves). 1. f3 was probably the more popular answer because it accomplishes nothing, weakens the king’s position, and takes away the king’s knight’s likely developing square. 1. g4, though more outrageously weakening, does prepare the bishop’s development, and against some defenses, may cramp Black’s game by a later advance to g5. There was a Dutch (I think, but maybe Danish) FM who had tried every legal opening move for White during his chess career with the single exception of 1. f3, because he couldn’t see a valid argument for it.
That seems rather odd to me as getting rid a central-ish pawn vs a wing pawn would certainly seem intuitive as causing more damage to the defensive structure of the side losing the pawn. I mean, getting rid of the f-pawn makes castling king side much less secure, it exposes a diagonal immediately to attack the king. It would have been among my first guesses, if not first guess, of which pawn’s removal would be the most advantageous to the side given odds.
There are some very early dangers to be aware of when playing without an f-pawn, for instance, if Black, giving the odds, replies to 1. e4 with …e5(??), he loses everything after 2. Qh5+. Instead, 1…Nc6 (preparing …e5) is a popular choice, and Philidor would often hunker down with 1…Nh6 followed by Nf7. Anyway, assuming Black avoid these traps and manages to castle kingside, his king can then hide in the corner, and the open f-file may give compensation.
Without a g-pawn, though, kingside castling is likely to be suicidal. You’re first of all castling onto an open file, and then the h-file will be full of undefended squares for your opponent to park his pieces on.
By the way, one of my shortest losing games with White since I made master was quite like an odds game. Black did admit to overlooking 11. Bxf7+. 34 years later, I’m still not up to examining it
closely but I’ll say that my play was routine and not mindful of the specific features of the position, particularly Black’s open f-file.
Appalling Gael–Bored Balkan